Updates on the Linux 2008-2012 guides

Since it’s been a while since I made an update on this guide and why it’s taking so long. Since it’s been so long since I’ve done any updates, here’s an update to explain the current situation. Hopefully this is the only major update in a while like this.

Wireless testing (and why I waited on Debian 9)

When I heard about Debian 9, I decided to wait so I don’t have to rush the information in and have implementation problems. I was able to get away with this at the time since Debian 9 was delayed so long and that would also delay LMDE 3. I could have chosen to not wait and put them out and amended them when Debian 9 came out and then updated the test notes, but I figured waiting and incorporating it right away would be better.

So far, this is where I stand on cards I have tested so far:

• 5 cards out of the group are half height, which are the easiest to test. These cards have already been tested in Debian 9.
• LMDE 3 has not been tested on any of the cards in the test pool yet (I have only tested Debian 9, Ubuntu 16.04.2 LTS, Linux Mint 18.1). It has not been released yet.
• Until LMDE 3 comes out, the full height cards will remain untested. It has to do with how hard the test hardware is to open.

As of now the half height cards are all fully tested with the exception of LMDE 3, but the full height cards will not be tested until LMDE 3 comes out and I will be able to publish the full test results (I may publish the testing notes for the half height cards, but not the full height cards). Other than full height cards and LMDE 3, the testing is done.

Debian 9 (and the changes on how firmware is installed if using a local file)

When it comes to Debian 9, I have also spent some time with it since I have the images available from the Wireless testing to give someone who doesn’t know an idea of what to expect. Debian 9 still has some known issues but for the most part, the bugs are fixed.

Debian 9 also changes how offline firmware installation is handled. Debian 9 no longer supports doing it through Synaptic Package Manger since the files get ran by the Software Center (which does not support firmware files). It has to be done though Terminal now, so this is just a heads up.

The hardware I am using for the guide (ThinkPad T420) has multiple issues

• Battery (runtime): There runtime on the battery is substantially worse than what it shipped with. The battery has wear and tear, but the problem is the gauge isn’t reliable since it isn’t accurately accounting for the remaining life of the pack. The worst that’s likely to happen is the pack will stop working to prevent an issue (but can be mitigated with higher low capacity warnings), so this is just an annoyance. It also takes longer to get to 100% then the pack is estimating it will take now, leading me to think it doesn’t have much actual life left (at least, used in a computer). At this point, I’ve accepted that the pack is never going to be accurate again.
• Battery (potential safety): The real issue with the battery is it has blinked like it has failed in the past, yet takes a charge (this happened when the pack was overdischarged). This no longer happens after it was recalibrated, but it affected the accuracy of the pack data and the reported runtime. I think the primary cause of the pack being damaged is the charger gave it 1A to charge when the pack was overdischarged, rather than charging it at a reduced current to start and then pushing full current when it was safer. I am okay with using it this way, but I will not be able to trust it is 100% safe anymore. (Replacement packs are usually ~$30 for a good aftermarket replacement).
• Ethernet: This looks repairable, since the only damaged part appears to be the jack. It has passed Intel diagnostics with flying colors, so I’m fairly certain the internal hardware isn’t damaged. The problem isn’t that it can’t be done; it’s just labor intensive and requires a full teardown to repair and is also soldered to the motherboard but I’m not against doing a board level swap to keep the parts cost down. (Jacks are $16/10; cost per jack is ~$1.50) (See revised repair costs)
• Headphone jack: Again, the part is cheap but the repair is labor intensive. Since it requires a full teardown, I’d rather do this while I repair the motherboard and get it over with. (Headphone jacks usually cost ~$8-34+ new/Lenovo FRU: 04W1623)
• Keyboard: Multiple keys have response issues where they do not respond the first time consistently. (Classic ThinkPad keyboards usually cost ~$30)

• Right now I have it to a point I feel like it is almost ready to be presentable. The original content from 2013-2014 is gone and the 2015 content where I tried to reuse the old content instead of starting over is almost gone, with the exception of 4 steps in the Laptop guide.
• Once I am done with the Laptop guide, I can start to consider making it public temporarily for feedback and work on the Desktop guide in the meantime. I copied the shared steps, and desktop unique steps to create a dedicated Desktop guide. I did this to better manage the content I am dealing with.
• I wish I didn’t reuse the old content by trying to save it. I feel like flagging for deletion and stating over would have been better.
• I am focusing more on "modular" content I can change to adapt, which is a lot easier to manage.
• I need to re-evaluate the system I picked and figure out if it’s worth repairing and if it will exceed 40% of the cost of a replacement system that doesn’t have major problems. If it meets or exceeds that threshold, the hardware is likely to be replaced. If it’s less, I may repair it. I am estimating it at $70 for everything I need to repair it.

Since I noticed the major issues with T420 thermal issues 2 days ago, I have been investigating the issue. I have tried the following things:

• Heatsink removal (dusting)
• New TIM

Neither of these fixed it, so I took the keyboard off to get an average on how hot the heatsink measures vs. what the computer reports back.

This is what I got from an IR thermometer:

• Vent: 57.6-67.8c
• Rear vent: 60c
• CPU plate: 61.2-66.6c
• Heatpipe: 35.2-52.4c

I have also measured it with a thermalcouple and did some datalogging and got these results:

CPU plate:

• 32-53c

Heatpipe 1:

• 42-49c

Heatpipe 2 (opposite side measurement)

• 42-48c

Now, what the computer reports

DOH! I miscalculated the repair costs from before. Here is a revised parts breakdown

Existing (will not change)

• Battery (two issues): Issue one: Runtime inaccuracy On the current battery I have, the runtime is substantially worse than what it gets when the battery is new. This is a wear and tear issue, so this would even happen to a new battery over time. The primary issue at play here is there is an issue with the battery controller being able to accurately track the actual remaining capacity of the cells and cannot accurately estimate the runtime anymore (and also invalidates 0-100 time estimates, since these are usually wrong as well). This is not a safety issue and can be worked around with higher low battery warnings. With that being said, the pack can get damaged or brick itself if it overdischarges because of this inaccuracy. I figured since this wasn't a safety issue, I could accept using the battery like this. Issue two: Potential safety At one point when I did not know about this runtime issue and assumed my recalibrations were good because they were accurate, I left the pack unplugged at night and went to sleep. Seeing as the pack was accurate when I recalibrated it, it should be fine and if it goes too low the controller will stop the boot and throw a 0190 error so I can plug it in, right? WRONG. This didn't happen and the pack overdischarged while I was sleeping one night. The pack did recover since I was able to recharge it, the light blinked like the pack had an internal error. Eventually this went away and the pack slowed down, so I assumed the pack just temporarily errored out. It continued to do this, even after this initial recovery. I was able to get rid of the light by recalibrating the battery, but this was done after it happened more then once.

All in all, the pack works now but the runtimen and data that's stored on the battery controller is no longer accurate. I also believe the cells were damaged since the computer charged a overdischarged pack with 1A, when it should have been smart enough to know to push reduced current into the battery before giving it the full 1A. I am okay with using the battery like this, but I'm prepared for it to report a permanent failure any day and will probably replace it and save it for a extra battery because of that. (A good aftermarket battery is ~$30)

• Headphone jack: The part is cheap but the repair is labor intensive. Since it requires a full teardown, I’d rather do this while I repair the motherboard and get it over with. (Headphone jacks usually cost ~$11-34+ new) Lenovo FRU: 04W1623
• Keyboard: Multiple keys have response issues where they do not respond the first time consistently. (Classic ThinkPad keyboards usually cost ~$30)

Revised

• Ethernet: I have a good feeling I can repair the motherboard, since the chipset passed Intel Diagnostics like nothing happened to the chipset that is part of the motherboard. With that being said, the part that I need is hard to find, since Lenovo decided to use a separate coil and jack design. What this means is I need the coils and an all in one Ethernet jack won't work to fix the issue. It appears the easiest way to get this part is another motherboard, but I don't want to waste a good motherboard for this coil. It looks like specialty shops sell the coil, but these shops are very hard to find. Of course, going external and considering the internal jack a lost cause also works but that's not the option I want to take if I don't have to.

My original points about the job being labor intensive stand (and my willingness to repair the motherboard to cut the costs stands). (Cost is TBD)

New

• The computer has developed a thermal issue. It is either the fan, CPU or heatsink at fault here. Right now, I need to see what parts cost and determine how expensive the repair can potentially get. I am going to try it without the dGPU enabled since I can switch this off, and see if that changed anything. (TBD)

I have a fan coming so I can begin troubleshooting the cooling problems on this ThinkPad and see if I'm going to need a heatsink. This is going to take some time to come, so the ThinkPad is out of use in the meantime.

I want to try a fan first to see if I will need a heatsink. If I need a heatsink that complicates the repair costs to a point I may not be able to economically repair the machine. The fan is only $7, so I don't mind trying it somewhat blind (and knowing it probably won't fix it). The heatsinks tend to be much more most of the time. I also want the high end heatsink, despite having the i5 (and in theory, I shouldn't need it). I want it more to keep the temps in check since the other cooler in the HMM has much less copper contact pad. It looks like it was designed for the i7, but that doesn't mean one can't install it on a i5 model. The points are the same, so it will fit.